IRC logs for #openttd on OFTC at 2023-06-14
⏴ go to previous day
00:51:57 *** Eddi|zuHause has quit IRC (Remote host closed the connection)
00:52:54 *** Eddi|zuHause has joined #openttd
02:51:18 *** debdog has quit IRC (Ping timeout: 480 seconds)
05:24:43 *** keikoz has quit IRC (Ping timeout: 480 seconds)
05:46:41 <pickpacket> You know what really sucks about boats? On a bigger map (I play on 1024x1024) they take more than a year to go two ways across one side of it. That means if I have an industry in one corner and ships taking cargo from it to a destination in the other corner on the same side they'll only turn a (huge) profit every other year
05:48:59 <EmperorJake> This is why you need daylength
06:28:17 <petern> Or use master with the changed cargo age rules...
06:45:46 <pickpacket> does any of that affect the boat speed?
06:49:14 <EmperorJake> But if a year is longer, the ships will be able to make many trips in a year at the same speed
06:55:57 <pickpacket> I don't know if I like the solution... Maybe I just feel like it's cheesing the system because I like to play within the existing limitations :)
06:56:46 <TrueBrain> But if we change the system, aren't that the new limitations, and as such not cheesing?
06:57:06 <EmperorJake> Fair enough, long daylength on JGR makes it a completely different game. But that could change in the future when economy and calendar length can be separated
06:58:39 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
07:19:18 <petern> I had that uninitialized issue before... and I can't remember how I solved it.
07:21:14 <petern> Although it was quite the same.
07:21:29 <petern> Okay my fix for 10938 is kinda ugly 😦
07:23:42 <TrueBrain> okay, new bananas infra is working with both old, current and new clients 😄 W00p! Now let's see if I can fix the DNS entry, and after that: let's move the production to this new infra .. will be fun .. what can go wrong 😛
07:24:46 <TrueBrain> let's first test if uploading still actually works 😛
07:26:23 <TrueBrain> hmm .. `if: ${{ github.actor != 'bananas-backend-staging' }}` didn't work ...
07:26:38 <TrueBrain> not a real problem, but .. that is meh .. what is the actor's name ...
07:31:40 <pickpacket> Jodie Foster? Bruce Willis?
07:37:30 <TrueBrain> now it actually works 🙂
07:37:57 <TrueBrain> although it crashes due to OOM .. but the preview is running on pretty low memory constraints, so that was kinda expected 😛
07:40:22 <TrueBrain> time to add 3GB of RAM to the cluster .. sadly, that also means the price goes up .. but hopefully I can scale down the old cluster a bit by the end of the day 🙂
07:42:35 <pickpacket> TrueBrain: how much does the price go up?
07:42:54 <TrueBrain> every doubling of RAM is about twice the price per instance
07:43:17 <pickpacket> What does that mean in numbers, though? :)
07:44:46 <pickpacket> I was planning to set up a monthly donation but discovered that liberapay requires you to transfer a sum to them first which will then be regularly distributed. So for me it doesn't really make a difference -- I still pay in a chunk
07:58:35 <TrueBrain> `s2n_init() failed: 335544378 (handshake was cancelled)` .. I love crazy errors, and asserts in AWS CLI tooling 😦
07:58:36 <andythenorth> github dependabot is picking up that old tags of grfs have vulnerable jquery in
07:58:51 <TrueBrain> andythenorth: lol 🙂
07:58:52 <andythenorth> do I retcon jquery and re-write each and every tag
07:59:07 <andythenorth> these html docs are deployed in prod, so the vuln is real
07:59:58 <andythenorth> I could undeploy the vulnerable tags
08:00:26 <andythenorth> a version of the docs is deployed for each tag
08:01:48 <andythenorth> might just delete the old versions
08:03:12 <petern> But also is it all just static sites? That probably limits the vuln?
08:03:43 <andythenorth> realistically yes
08:04:07 <petern> Crap, it's past 9am already 😦
08:04:33 <andythenorth> there are 2 vulns, both XSS
08:05:04 <petern> Of course, jquery is old enough that any new release is probably a security fix...
08:05:05 <andythenorth> I don't do any XSS, and I don't accept user input
08:05:24 <andythenorth> jquery is only used because Bootstrap needs it for the dropdown menu on small viewports 😛
08:05:35 <andythenorth> but dependabot requires a fix, so eh
08:05:39 <andythenorth> scorch the earth
08:05:49 <andythenorth> deleting tags used to be considered vandalism?
08:05:50 <petern> Yeah, IIRC bootstrap 5 removes the jquery requirement?
08:05:57 <petern> But that's also another upgrade...
08:05:59 <andythenorth> jquery might be dead soon
08:06:20 <andythenorth> maybe I just fork the repos, make a new tag, and delete everything else
08:06:23 <petern> Unlikely. There's a v5 in progress. (Skipping v4 I guess?)
08:06:33 <andythenorth> because the vulnerable jquery will be in all the git reflogs
08:06:39 <andythenorth> so still technically fails security
08:06:49 <andythenorth> unless I manually edit all the reflogs
08:10:26 <pickpacket> andythenorth: which website are you talking about now?
08:11:48 <petern> Argh, Americans in videos talking about chassis...
08:26:01 <pickpacket> petern: always salad?
08:31:02 *** greeter has joined #openttd
08:33:23 <petern> But urgh, really hungry for some reason. Time for a breakfast banana.
08:42:37 <TrueBrain> ugh ... cgroups are having fun ... after a while, I can't access `/dev/urandom` or `/dev/null` anymore .. known bug in Nomad for over a year .. kinda essential, to have things working constantly 😄
08:46:07 <petern> cgroups are 16 years old... oh.
08:46:23 <TrueBrain> yeah, but v2 is giving a lot of issues where ever you look 😛
08:51:29 <petern> Ooh, there's a new thing for me to get wrong 😄
08:53:09 <TrueBrain> lol, guess what ... AWS' container registry .. IPv4 only 😄
08:55:09 <TrueBrain> so how am I suppose to work around this cgroup issue ........ hmmmmm
08:57:32 <TrueBrain> it is so weird it works for a while, then stops working .. I don't like systems that act like that
08:58:09 <SpComb> AWS would lose 10% of revenue if everybody stopped setting up NAT Gateways for their VPC private subnets
08:58:32 <TrueBrain> the amount they charge for a NAT Gateway has always been ......... weird
09:06:50 <pickpacket> hm. I wonder if that's something cloudflare's argo tunnel could do instead
09:06:53 <TrueBrain> anyway ... I did not want to spend over an hour on stupid cgroups .. I wanted to start bringing bananas over to the new infra 😛
09:08:03 *** Deep3D has quit IRC (Ping timeout: 480 seconds)
09:08:15 <pickpacket> whut? That's supposed to be a gif
09:08:20 <TrueBrain> pickpacket: no; Cloudflare Tunnel isn't a NAT gateway
09:09:22 <pickpacket> at previous job we used cloudflare tunnel to avoid exposing public IPs in GCP. I guess that's a different thing
09:09:36 <TrueBrain> it is a tunnel, to get things from A to B 🙂
09:09:45 <TrueBrain> not from A to the whole world
09:09:57 <TrueBrain> both A and B are pretty well defined
09:10:29 <pickpacket> tbf I don't know what's behind the NAT gateway you're talking about :D
09:11:38 <pickpacket> I have sooo much on my schedule today. Both bad, neutral, and good stuff. But honestly I just want to go home and play OpenTTD
09:22:01 <TrueBrain> right, as for this migration ... how am I going to do that .. hmmmmm
09:24:19 <TrueBrain> using fake domains first, I guess, so I can migrate all data etc etc
09:32:59 <TrueBrain> migrating all BaNaNaS files .. our AWS bill might become a bit higher this month 😛
09:34:12 <TrueBrain> could be worse .. it could have been our CDN 😄
09:34:12 <CK2347> who pays the bills though
09:34:36 <TrueBrain> we have awesome people who donate money to our project 🙂
09:34:38 <TrueBrain> keeps the lights on
09:35:03 <CK2347> I though owen took care of this as well
09:35:18 <TrueBrain> you think he is bankrolling this project? 😒
09:35:34 <TrueBrain> he does track our donations, and make sure all our bills are paid, if that is what you meant 🙂
09:36:28 <TrueBrain> our BaNaNaS bucket is 18GB in size .. I am always surprised by that number 😛
09:36:33 <CK2347> like you are ( I guess)
09:37:05 <CK2347> TrueBrain: That's not much for openttd ,right?
09:37:06 <TrueBrain> this will cost is ~27 dollarcent per month, to store that! 😄 (Cloudflare prices are so much more reasonable than AWS)
09:37:21 <TrueBrain> I think 18GB of trains graphics is a shitload
09:37:36 <CK2347> Also does old stuff from the bananas gets removed?
09:37:47 <TrueBrain> savegame compatibility and all
09:37:51 <TrueBrain> I think it really shouldn't
09:41:56 *** Kitrana1 has joined #openttd
09:46:59 *** Kitrana has quit IRC (Ping timeout: 480 seconds)
09:48:16 <pickpacket> as in $0.27 per month for 18GB of storage?
09:49:42 <TrueBrain> ideally if Cloudflare ever got back to me about sponsorship, it might be even cheaper. But I just don't know if they are still looking at it, ignored it, or that it was never filed properly .. *shrug*
10:00:25 *** D-HUND is now known as debdog
10:01:15 <Eddi|zuHause> my steam developed this habit of losing connection, and then not reconnecting until a client restart...
10:22:02 <petern> Oops, I was meant to post on Facebook :/
10:22:14 <TrueBrain> wait, Facebook still exists?
10:29:01 <Eddi|zuHause> at the current rate, twitter is going the way of the facebook
10:29:15 <TrueBrain> reddit is playing catchup
10:29:29 <petern> Okay, slightly happier that it wasn't me introducing broken things... just not knowing the existing stuff was broken and continuing it...
10:29:49 <petern> Funny how years-broken stuff can apparently work fine 🙂
10:31:27 <Eddi|zuHause> that's neither a heisenbug or bohrbug... what's it?
10:32:26 <Eddi|zuHause> it's like two bugs cancelling each other out
10:52:08 *** _aD has quit IRC (Ping timeout: 480 seconds)
11:21:33 <Eddi|zuHause> cities:skylines is running unusually fast... something must be wrong
11:27:03 *** _aD has quit IRC (Ping timeout: 480 seconds)
12:34:28 <TrueBrain> okay, in theory I can migrate over bananas .. all pre-checks show everything should "just work" .. but that cgroup issue in the cluster needs addressing first, I guess ..
12:34:33 <TrueBrain> as that can cause downtime when I am not around 🙂
12:37:58 <TrueBrain> just no clue yet how to resolve that ... meh
12:39:14 <TrueBrain> glx[d]: would you happen to have some time and can you test IPv6 to the new infra for me? `OTTD_CONTENT_MIRROR_URI="https://bananas-preview-server.openttd.org/bananas" OTTD_CONTENT_SERVER_CS="content-preview.openttd.org:4978" ./openttd -dnet=6` is the Linux variant, something similar. It should show very few entries in the online content service, and you should be able to download them
12:40:51 <TrueBrain> this is for master, the above command
12:40:56 <TrueBrain> for 13 and below it is slightly different
12:41:05 <TrueBrain> so depends on what you prefer 😄
12:41:36 <glx[d]> steam in nightly mode should do it then 🙂
12:41:52 <TrueBrain> no clue if and how you can change env-variables in Steam 😛
12:42:05 <glx[d]> I'll run it from powershell
12:43:11 <TrueBrain> even if you quote those URLs, there is always some client (like the lounge client) that polls the URL
12:43:18 <TrueBrain> at least I now have a list of IPs of people using lounge 😛
12:43:44 <TrueBrain> URL previews is so bad for privacy, when not routed via a global entity ..
12:46:14 <TrueBrain> downloading works too? And it is over IPv6?
12:46:32 <glx[d]> dbg: [net] Attempting to connect to [2a05:d018:5fd:4605:409b::1]:4978 (IPv6)
12:46:32 <glx[d]> dbg: [net] Connected to content-preview.openttd.org:4978
12:46:32 <glx[d]> dbg: [net] - using [2a05:d018:5fd:4605:409b::1]:4978 (IPv6)
12:46:43 <TrueBrain> awesome! At least something that goes right first time with IPv6 🙂
12:47:04 <TrueBrain> hmm, that seems to be a download over TCP, not over HTTP 😛
12:47:17 <glx[d]> dbg: [net] HTTP request to bananas-preview-server.openttd.org/bananas
12:47:17 <glx[d]> dbg: [net] HTTP request failed: L’adresse ou le nom de serveur n’a pas pu être résolu
12:47:29 <glx[d]> looks like I failed somewhere maybe
12:48:15 <TrueBrain> ah 🙂 Yeah, that part is kinda mandatory 😛
12:49:09 <petern> Meh, it used ipv4 for me.
12:49:20 <glx[d]> better with the correct URO
12:49:39 <glx[d]> many `dbg: [net] HTTP callback: 8192 bytes`
12:50:04 <TrueBrain> petern: via WSL it always does, sadly 😛
12:50:56 <glx[d]> dbg: [net] Attempting to connect to [2a05:d018:5fd:4604:dbc0::1]:4978 (IPv6)
12:50:56 <glx[d]> dbg: [net] Connected to content-preview.openttd.org:4978
12:50:56 <glx[d]> dbg: [net] - using [2a05:d018:5fd:4604:dbc0::1]:4978 (IPv6)
12:50:56 <glx[d]> dbg: [net] HTTP request status code: 200
12:50:57 <glx[d]> dbg: [net] HTTP callback: 138 bytes
12:50:57 <glx[d]> dbg: [net] HTTP callback: 0 bytes
12:51:06 <glx[d]> hard to tell for the https part
12:51:22 <TrueBrain> ah, WinHttp doesn't tell you .. well, should be fine 🙂
12:52:07 <glx[d]> but my connection is native IPv6 so it should default to IPv6
12:52:22 <TrueBrain> yeah, but not that it did a fallback to IPv4 for what-ever reason; but this looks good 🙂
12:53:06 <glx[d]> I could check via ressource monitor too
12:53:26 <TrueBrain> you accidentially also tested that TCP transfers are IPv6 .. so this just looks good 🙂
12:53:50 <TrueBrain> so now all I have to battle is this cgroup issue .. and we should be good to go \o/
12:55:14 <TrueBrain> Cloudflare is the entry for most connections .. only custom TCP goes in via AWS (as Cloudflare only does HTTP .. at least, for now :P)
12:55:28 <TrueBrain> and the difference in IPv6 prefixes show that nicely 😄
13:02:15 <glx[d]> oh indeed wsl is annoyingly ipv4 only
13:03:25 <glx[d]> WSL 2 distributions currently cannot reach IPv6-only addresses. We are working on adding this feature.
13:22:46 <TrueBrain> it isn't easy adding that .. as it kinda depends what kind of IP the host got
13:22:52 <TrueBrain> you don't want to NAT IPv6 traffic
13:23:01 <TrueBrain> so I don't expect this to happen any time soon 😛
13:35:41 <TrueBrain> DHCPv6 is possible, but still depends on a node in your network that supports that
13:36:06 <TrueBrain> if your IPv6 is configured correctly, you should get a /64 assigned to your computer .. but this isn't always the case .. some providers only give a /64 to the router itself
13:36:20 <TrueBrain> so not sure how they will solve it for WSL 🙂
13:38:24 <TrueBrain> a /48 or a /56 should be given to you by your provider 🙂
13:38:28 <TrueBrain> but some are .. cheap
13:38:38 <TrueBrain> like .. "BUT WHAT IF IPv4 REPEATS ITSELF" cheap
13:38:48 <TrueBrain> they don't actually understand those / numbers 😛
13:38:50 <petern> And lots of routeres are shitty enough to no have any configuration for these things.
13:39:20 <petern> I have a few /64s allocated to different VLANs, like a normal home user...
13:39:37 <TrueBrain> well, a single machine should receive a /64 if done properly 😛
13:39:44 <TrueBrain> so a VLAN should be more like a ... /56 ? 😄
13:40:09 <TrueBrain> can't even remember what is assigned to my machine by my router .. let's check ..
13:41:43 <TrueBrain> ah, yes, a /64 per "VLAN" I did here too 🙂
13:41:51 <TrueBrain> it's ffiiinnnneeeeee
13:43:56 <TrueBrain> meh; this cgroup issue is annoyyyyyiiiiiinnnngggggg
13:49:05 <petern> Yeah no, not a whole /64 per machine.
13:49:32 <TrueBrain> on AWS a single machine gets a /80 🙂
13:49:40 <TrueBrain> like .. that is a lot 😛
13:50:31 <Eddi|zuHause> "a /64 ought to be enough for anyone"? :p
13:53:44 <petern> I also have a /29 which I've allocated as 8 /32s to avoid wastage.
13:53:52 <petern> Perfectly normal home-user stuff.
13:54:04 <TrueBrain> do I want to know why you have a /29 at home? 😛
13:54:39 <petern> For my servers of course.
13:56:13 <TrueBrain> `failed to open '/dev/zero': Operation not permitted`
13:56:16 <TrueBrain> it is just .. weird
13:56:27 <TrueBrain> and I don't understand cgroups enough to say anything useful about this
13:57:23 <TrueBrain> mostly what puzzles me, is why this happens after N minutes, and not directly
13:58:27 <Eddi|zuHause> your free usage of /dev/zero has been used up, upgrade to a premium subscription of /dev/zero for 4,99$
13:58:43 <TrueBrain> everything above (or below, depends how you look at it) a `/64` for home usage is fine 🙂 So a /60 is more than fine
13:59:49 <glx[d]> I can even configure next hops
14:00:02 <glx[d]> inside the ISP provided router
14:01:33 <glx[d]> dunno it's for prefix delegation on different /64 subnets
14:02:21 <glx[d]> I guess some users use that
14:04:43 <glx[d]> the only "advanced" feature I use is the fullstack IPv4 (1/4 IPv4 is a pain)
14:26:05 <TrueBrain> okay .. so most things in cgroups you can control via the filesystem
14:26:09 <TrueBrain> but the device controller works with BPF
14:36:12 <TrueBrain> okay .. for some reason, at some point in time, the BPF is changed
14:36:15 <TrueBrain> like .. totally changed
14:40:01 <petern> > 2312 packages can be upgraded. Run 'apt list --upgradable' to see them.
14:44:33 *** audunm[m] has joined #openttd
14:45:13 *** audunm[m] has left #openttd
14:46:07 <pickpacket> Is Bullseye number 11?
15:14:58 <TrueBrain> right, I think I found the issue of the cgroup problems ... when ever a docker is started on the host, all BPFs of the existing cgroups are overwritten with a "deny everything" one
15:15:14 <TrueBrain> so things work, till a docker wants to reboot
15:15:16 <Eddi|zuHause> that sounds helpful
15:15:35 <TrueBrain> no, not really. At least it might show how to prevent the problem from happening, at least for now 🙂
15:17:42 *** gelignite has joined #openttd
15:18:25 <Eddi|zuHause> "never reboot docker"?
15:18:33 <TrueBrain> never reboot any container
15:18:35 <TrueBrain> which is impossible
15:18:39 <TrueBrain> things break, stuff crashes
15:26:37 <petern> I like your modest "I don't know what I'm looking at, but there's why it's happening"
15:27:51 <TrueBrain> it was fun to learn how the fuck cgroup and BPF worked 🙂
15:28:42 <TrueBrain> so okay, you can only have one active `exec` driver on Nomad, and it better be the last one that started 😄
15:33:35 <glx[d]> seems more fun than compiler bugs
15:33:38 <TrueBrain> looking at the depth of this problem, I cannot imagine that Nomad actually runs fine with cgroupv2, if you have anything except dockers running ...
15:33:56 <TrueBrain> well, if you don't use anything from `/dev` you will be fine, I guess ..
15:34:04 <TrueBrain> so no `urandom` for you! 🙂
15:41:01 *** Eddi|zuHause has quit IRC (Remote host closed the connection)
15:42:56 *** Eddi|zuHause has joined #openttd
16:07:10 *** _aD has quit IRC (Quit: leaving)
16:38:27 *** HerzogDeXtEr has joined #openttd
17:29:24 <orudge> TrueBrain: Good news, I got my OS/2 environment up and running again! It looks like I have a diff from last time with some OS/2 fixes, I wonder if it'll still work :D
17:29:59 <TrueBrain> I still really wonder if it is worth keeping the technical debt in our code for ... the once every 2 years you boot your machine 😛
17:30:36 <orudge> I don't know that I am necessarily inclined to keep it around, but I'd like to test it out first and see, more out of curiosity than anything else
17:31:01 <TrueBrain> knock yourself out 🙂
17:52:41 <glx[d]> we try to spot potential compile issues when OS/2 stuff is touched, but it's only by reading the PR
18:48:53 <TrueBrain> haha, yes, survey key also needs a check, and just skipped over 🙂
18:49:17 <TrueBrain> orudge: without those checks, it becomes impossible to thinker with the release pipeline for anyone else .. so yeah, we should try to make them work even without any secrets 🙂
18:49:39 <TrueBrain> but yeah, if someone can also slap an if-statement around the survey-key, would be lovely 🙂
19:03:20 <TrueBrain> Found it! But the problem is ........ there is no shortcut to fix it without them accepting the PR and making a release 😦
19:09:31 <Rubidium_> do I see aligned parameters there? :D
19:09:44 <TrueBrain> yeah ... Go coding style ... and it shows everything I hate about that
19:09:58 <TrueBrain> so you do have a trailing comma
19:10:01 <TrueBrain> and STILL you align?
19:10:05 <TrueBrain> it is like ... seriously .....
19:10:09 <TrueBrain> and also: fuck you Rb 😛
19:11:48 <Rubidium_> didn't want to make you mad :(
19:13:21 <JGR> On the upside, gofmt will do all that busywork for you
19:13:32 <TrueBrain> yup; didn't even type the space
19:13:54 <JGR> And you don't have to argue about its configuration because there basically isn't any 😛
19:14:00 <TrueBrain> but boy, I never worked with Go, and I already just dislike how it is structured ... in most cases it took me a long time to find out where something came from, which package supplied it, or where it goes to ..
19:14:18 <TrueBrain> JGR: same reason I love `black` in Python. One opinion. Done.
19:16:02 <JGR> I've done a fair bit of go at work. It feels a bit limiting but that is not necessarily a bad thing with large teams.
19:16:32 <JGR> Green threads whch aren't terrible is also very nice.
19:17:15 <TrueBrain> seems the only way to get around the problem I am facing, is removing cpuset from cgroup in the kernel cmdline .. but that is a bit tricky to do on AWS, from what I can tell ..
19:18:21 <TrueBrain> so no new and faster BaNaNaS yet 😦
19:24:09 *** esselfe has quit IRC (Remote host closed the connection)
19:30:14 *** esselfe has joined #openttd
19:46:09 <frosch> eints has so much legacy code...
19:47:18 <frosch> 26 files changed, 784 insertions(+), 2169 deletions(-)
19:47:36 <frosch> for removing xml and json storage types, and adding sqlite instead
19:47:52 <frosch> most of it is deleting xml stuff
19:49:11 *** gelignite has quit IRC (Quit: Stay safe!)
21:00:45 <glx[d]> was testing {COLOUR} after a rebase, seems {RAW_STRING} has some problems
21:02:57 <glx[d]> oh it's my fault, I approved the change
21:29:28 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:00:32 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
22:12:30 *** audunm[m] has joined #openttd
22:25:31 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
continue to next day ⏵
